humdrumR 7.0.7
Fixed a bugs in timeline(), so it is now working, though I still hasn’t been extensively tested and I consider it “under development.”
Also fixed a few other minor bugs.
humdrumR 7.0.6
We’ve fixed some minor bugs which were causing major problems in important functions:
- There was a bug happening if you tried to used
filter()(orsubset()) twice on the same data. This should be fixed now. - There were several bugs in
metcount()leading to errors. These should also be fixed.
While I was at it, I’ve added .humdrumR() methods for a bunch of rhythm functions (metcount.humdrumR(), metsubpos.humdrum(), metlev.humdrumR(), etc.) so that these functions can be used directly on humdrumR data objects. For example:
chorales <- readHumdrum(humdrumRroot, 'HumdrumData/BachChorales/.*.krn')
chorales |> metcount(level = '16')
Speaking of metcount(), I also made a change so that it counts the given beat level inside the full span of the meter, instead of always within the next highest level. This is much more intuitive in most cases. You can still get the old behavior using a new argument, withinNext = FALSE.
humdrumR 7.0.5
Version 7.0.5 includes a small patch to fix a bug related to the :: function, as well as a new feature/behavior which builds off of that fix.
In previous versions, an error would occur if you used :: inside any of the fundamental humdrumR “with/within” methods (including base functions and dplyr “verbs”). We’ve fixed this bug.
We’ve also added a new feature; specifically, we’ve made using :: in humdrumR less necessary. Now, within a humdrumR call to any of these methods (listed below), if you use any function exported by humdrumR, humdrumR will automatically use the humdrumR version of that function, even if another package you have attached includes a function with same name. In other words, the humdrumR namespace will always takes priority within a humdrumR method call.
For example, the dplyr package exports a function called lag() and so does humdrumR (these functions do the same thing, but humdrumR’s version has a few extra features). Before this update, if you loaded dplyr after you loaded humdrumR, then dplyr’s lag() function would generally take priority over humdrumR’s lag(). So, code like this
library(humdrumR)
library(dplyr)
readHumdrum(humdrumRroot, 'HumdrumData/BachChorales/.*krn') -> chorales
chorales |> mutate(Lagged = lag(Token))
would call dplyr::lag(). This sort of behavior can be confusing but wouldn’t normally be the end of the world (this is how R is supposed to work, after all). However, the lag() function is particularly problematic because many humdrumR methods rely on our special version of lag(). This is why we’ve implemented a change: Now, if you use lag() within a humdrumR with/within method, it will default to using the humdrumR version, regardless of what other packages are loaded. So the code above would use humdrumR::lag(). If you want to use dplyr’s version (or any other package), you still can by specifying (for example) dplyr::lag(). Another function which (in the past) could lead to frequent namespace confusion was transpose()—now, you can safely use transpose() and know that your system will use humdrumR::transpose().
All .humdrumR methods for the following functions are affected by this change:
humdrumR 7.0.3
count() and table()
We wanted to use the tally() as humdrumR’s main “count stuff” function. Unfortunately, we found that tally() already exists in dplyr and it (the generic tally) can’t really be extended as we’d like. We’ve reimplemented everything focusing on an extension of the dplyr::count() function. This also necessitated renaming a our (metric) count() function to timecount(). Our count() method now generates a cool “distribution” table, which is easily to manipulate. This is all mostly working pretty well, but is still a work in progress and not documented.
We’ve also extended base::table() to work with humdrumR data, but we’ll be emphasizing using count() in our documentation.
humdrumR 7.0.0
HumdrumR 0.7.0.0 includes some pretty major changes from the previous (0.6.x.x) versions. Notably, we have updated the whole package to make it more consistent with the Tidy-verse. We’ve also incorporated an option to use humdrumR functions in a super-concise manner which is closely modeled off the original humdrum toolkit. Finally, we have changed humdrumR data objects to behave a little more like “normal” R data.frames (or tibbles), and allowed you to more easily “see” and manipulate your data as a data.frame. Our reference manuals and articles have all been updated to reflect these changes.
Tidy-verse style
HumdrumR has been changed in a few ways align more with the Tidy-verse, and particular, the dplyr package. The following dplyr packages now have humdrumR methods:
select()filter()mutate()reframe()summarize()group_by()pull()
select()
The biggest change to the package is the use of the select() function. In humdrumR, we have dropped the use of term “Active Field” in favor of the term “Selected Field.” Use the select() function to select fields—you can no longer use $ for this purpose.
mutate(), summarize(), and reframe()
We now use the dplyr “verbs” mutate(), summarize(), and reframe() as our main tools for manipulating humdrumR data. These commands work very similarly to the within() and with() functions from previous humdrumR versions, but are slightly different in how they recycle results. The base-R with() and within() functions can still be used as before—in fact, with() is still particularly useful sometimes, because it is basically the same as going reframe() |> pull() in one step.
group_by()
Instead of providing a by = argument to humdrumR functions, you should now call group_by() as a separate command. Your data will stay grouped until you call ungroup().
pull()
The dplyr “verb” pull() can also be used to extract a field from a humdrumR dataset. There are also functions pull_data.frame(), pull_data.table() (requires data.table, and pull_tibble() (requires tibble), to pull multiple fields into a “normal” R data.frame/data.table/tibble.
Note that the $ command has been changed to be consistent with base-R and the Tidyverse. This command will now simply extract a field from humdrumR data.
Humdrum style
A new feature is that many of humdrumR’s functions can be applied directly to humdrumR data, without using mutate(), within(), etc. For example, you can take some humdrum data and pipe it directly to the kern() function, like: humData |> kern().
Data.frame view
A new option in humdrumR 0.7.0.0 is to view humdrumR data a normal R data.frame, instead of showing the data as a humdrum-syntax score. Call the command humdrumR("data.frame") once to switch to the new view. When in “data.frame view,” humdrumR will still only show you the selected fields. If you want to see all the fields, use select(everythign()).
If you want to switch back to the “humdrum view” (the default), call humdrumR("humdrum").